OCR Search / Search in images, PDFs, etc

Feature Requests
,
1

Is your feature request related to a problem? Please describe.

I like to have textpassages of writings or a text of a book in my notes but don’t like to type it myself.

And beeing able to search trhough memories, like photographed birthdaycards.

Describe the solution you’d like

Inbuild OCR to right-click on pictures and choose ‘OCR’-it.

Friends keep telling me of Evernote in that sense.

Describe alternatives you’ve considered

using a third party app, e.g. Camscanner or Evernote and sending the note over. but… no :smiley: in terms of dataprivacy no valid option.

3

feature request

Search needs some big update. One of which is Find Text from images and pdf documents.

Solution

Find text from images and pdf like how evernote does.

evernote-search-within-pdf-files-1024x634
evernote-search-within-pdf-files-1024x6341024×634 235 KB

4

+1 for this. Really handy for making notes

5

I think this is a highly underrated feature that OneNote offers. The combination of OCR and good search brings so much value!

6

Hello everyone :nerd_face: Thanks for rising this up. I will add this idea in our product backlog.
For now Anytype is not really optimised for working with visual information like pdf and images. And we want address this request as a part of integral initiative “visual content as a first class citizen” to provide solid experience.
For now we are actively preparing for public release and we could start this initiative after public release.

7
  • 1 for the feature. My use for Evernote is primarily because of if it’s excellent search capabilities especially in the pdfs.
Majona 8

:+1:t4:

9

OCR would be great: +1.

But as a first step I’d be quite satisfied for Anytype to search the contents of PDFs that already have a text-element layer: +11.

:slight_smile:

10

+1, Evernote is quite impressive!

11

+1 from me. OCR would be huge :slight_smile:

12

11/8/23
This was probably mentioned before, but OneNote also has this hability, and it seems to process the images once at their creation, and keep the text assosiated on the background, so the search doesn’t have to trigger the OCR each time. Also, getting the word highlighted is a must. I know this is a big feature, but I think it’s one that most people would benefit from.

Edit 3/3/25:
Just came back to see if there was any news regarding this request. I’ll add that this other request would complement OCR capability quite nicely: Mobile document scanner
I also can’t believe this request is +3 years old. They grow up so quickly :rofl:

13

Hello, may I ask if this feature will be implemented?Search for text on the image.

14

Implementing this is definitely challenging because Tesseract is quite outdated and fails about 99% of the time. There are much better vision models available here: Models - Hugging Face. However, integrating a high-quality, compact vision model into Anytype seems almost impossible. While it would be a game-changer and incredibly impressive, I’m not very optimistic about seeing a reliable OCR feature implemented anytime soon.

15

And they make it so easy too. Right click → “copy text from picture”
This is awesome functionality and very helpful - would be good to have it here.

16

Here is an open source OCR model